Binding media stream audio track to speech recognition [renderer]

content/renderer/media/speech_recognition_audio_sink.h

Index: content/renderer/media/speech_recognition_audio_sink.h
diff --git a/content/renderer/media/speech_recognition_audio_sink.h b/content/renderer/media/speech_recognition_audio_sink.h
new file mode 100644
index 0000000000000000000000000000000000000000..86605c420001391b22e82d90070d08090fddba40
--- /dev/null
+++ b/content/renderer/media/speech_recognition_audio_sink.h
@@ -0,0 +1,128 @@
+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef CONTENT_RENDERER_MEDIA_SPEECH_RECOGNITION_AUDIO_SINK_H_
+#define CONTENT_RENDERER_MEDIA_SPEECH_RECOGNITION_AUDIO_SINK_H_
+
+#include "base/callback.h"
+#include "base/memory/scoped_ptr.h"
+#include "base/memory/shared_memory.h"
+#include "base/sync_socket.h"
+#include "base/threading/thread_checker.h"
+#include "content/common/content_export.h"
+#include "content/public/renderer/media_stream_audio_sink.h"
+#include "content/renderer/media/media_stream_audio_source.h"
+#include "media/audio/audio_parameters.h"
+#include "media/base/audio_converter.h"
+#include "third_party/WebKit/public/platform/WebMediaStreamTrack.h"
+
+namespace media {
+class AudioBus;
+class AudioFifo;
+}
+
+namespace content {
+
+// SpeechRecognitionAudioSink works as an audio sink to the
+// WebRtcLocalAudioTrack. It stores the capture data into a FIFO.
+// When the FIFO has enough data for resampling, it converts it,
+// passes the buffer to the WebSpeechRecognizer via SharedMemory
+// and notifies it via SyncSocket followed by incrementing the |buffer_index_|.
+// WebSpeechRecognizer increments the shared buffer index to synchronize.

[henrika (OOO until Aug 14), 2014/09/29 10:38:41]

Can you clarify? What is synchronized and how does it work?

[burnik, 2014/09/29 12:07:31]

The buffer indices are synchronized.
Detailed in design doc. http://goo.gl/9Ot3PC

+class CONTENT_EXPORT SpeechRecognitionAudioSink
+    : NON_EXPORTED_BASE(public media::AudioConverter::InputCallback),
+      NON_EXPORTED_BASE(public MediaStreamAudioSink) {
+ public:
+  typedef base::Callback<void()> OnStoppedCB;
+
+  SpeechRecognitionAudioSink(/* ExtraData reference is copied from track. */
+                             const blink::WebMediaStreamTrack& track,
+
+                             /* Output audio parameters are copied. */
+                             const media::AudioParameters& params,
+
+                             /* Handle is used to map to existing memory. */
+                             const base::SharedMemoryHandle memory,
+
+                             /* Socket ownership is passed to here. */

[henrika (OOO until Aug 14), 2014/09/29 10:38:41]

'to here' sounds odd. Can you rewrite?

[burnik, 2014/09/29 12:07:31]

Done:
*Socket ownership is transferred.*

+                             scoped_ptr<base::SyncSocket> socket,
+
+                             /* Callback is stored by copy. */
+                             const OnStoppedCB& on_stopped_cb);
+
+  virtual ~SpeechRecognitionAudioSink();
+
+  // Returns whether the provided track is supported.
+  static bool IsSupportedTrack(const blink::WebMediaStreamTrack& track);
+
+ private:
+  // content::MediaStreamAudioSink implementation.
+  virtual void OnReadyStateChanged(
+      blink::WebMediaStreamSource::ReadyState state) OVERRIDE;
+
+  virtual void OnData(const int16* audio_data, int sample_rate,
+                      int number_of_channels, int number_of_frames) OVERRIDE;
+  virtual void OnSetFormat(const media::AudioParameters& params) OVERRIDE;
+
+  // media::AudioConverter::Inputcallback implementation.
+  virtual double ProvideInput(media::AudioBus* audio_bus,
+                              base::TimeDelta buffer_delay) OVERRIDE;
+
+  // Number of frames per buffer in FIFO. When the buffer is full we convert and
+  // consume it on the |output_bus_|. Size of the buffer depends on the
+  // resampler. Example: for 44.1 to 16.0 conversion, it should be 4100 frames.
+  int fifo_buffer_size_;
+
+  // Used to DCHECK that some methods are called on the main render thread.
+  base::ThreadChecker main_render_thread_checker_;
+
+  // Used to DCHECK that some methods are called on the capture audio thread.
+  base::ThreadChecker capture_thread_checker_;
+
+  // The audio track that this audio sink is connected to.
+  const blink::WebMediaStreamTrack track_;
+
+  // Shared memory used by audio buses on both browser and renderer processes.
+  base::SharedMemory shared_memory_;
+
+  // Socket for synchronization of audio bus reads/writes.
+  // Created on the renderer client and passed here. Accessed on capture thread.
+  scoped_ptr<base::SyncSocket> socket_;
+
+  // Used as a resampler to deliver appropriate format to speech recognition.
+  scoped_ptr<media::AudioConverter> audio_converter_;
+
+  // FIFO is used for queuing audio frames before we resample.
+  scoped_ptr<media::AudioFifo> fifo_;
+
+  // Audio delivered from source.
+  scoped_ptr<media::AudioBus> input_bus_;
+
+  // Audio bus shared with the browser process via |shared_memory_|.
+  scoped_ptr<media::AudioBus> output_bus_;
+
+  // Params of the source audio. Can change when |OnSetFormat()| occurs.
+  media::AudioParameters input_params_;
+
+  // Params used by speech recognition.
+  const media::AudioParameters output_params_;
+
+  // Whether the track has been stopped.
+  bool track_stopped_;
+
+  // Local counter of audio buffers for synchronization.
+  uint32 buffer_index_;
+
+  // Peer's counter of audio buffers for synchronization.
+  const uint32* peer_buffer_index_;
+
+  // Callback for the renderer client. Called when the audio track was stopped.
+  const OnStoppedCB on_stopped_cb_;
+
+  DISALLOW_COPY_AND_ASSIGN(SpeechRecognitionAudioSink);
+};
+
+}  // namespace content
+
+#endif  // CONTENT_RENDERER_MEDIA_SPEECH_RECOGNITION_AUDIO_SINK_H_